Feel like an Artist

The Mart, Museum of Modern and Contemporary Art of Trento and Rovereto, has a collection of 20,000 images of works of art by Italian and international artists. We have developed a chatbot that offers fun, innovative and immediate access to a part of this heritage.

Our idea was to have the users describe their emotions, and then to discover works of art by artists who may have experienced similar moods.

How did we do it? We collected about a hundred sentences containing emotional states to calibrate a bipolar (negative / positive) sentiment analysis classifier. This way, we created a scale of intensity of the emotional states (neutral, very_low, low, high, very_high).

We then asked a cultural heritage communication expert to tag 50 images.

"Feel like an artist" therefore returns personalized access to quality content - which can also become a stimulus for visiting a cultural institution and perhaps, in the not too distant future, an effective way to interact with an online database of collections.

Built With

• chatfuel
• elasticsearch
• love
• python
• umbraco